CN111815597A - Left ventricle long and short axis tangent plane extraction method and device based on CT image, computer equipment and storage medium - Google Patents

Left ventricle long and short axis tangent plane extraction method and device based on CT image, computer equipment and storage medium Download PDF

Info

Publication number
CN111815597A
CN111815597A CN202010623486.XA CN202010623486A CN111815597A CN 111815597 A CN111815597 A CN 111815597A CN 202010623486 A CN202010623486 A CN 202010623486A CN 111815597 A CN111815597 A CN 111815597A
Authority
CN
China
Prior art keywords
left ventricle
long
image
axis
mitral valve
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010623486.XA
Other languages
Chinese (zh)
Other versions
CN111815597B (en
Inventor
潘轶斌
何京松
李长岭
向建平
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Arteryflow Technology Co ltd
Original Assignee
Hangzhou Arteryflow Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Arteryflow Technology Co ltd filed Critical Hangzhou Arteryflow Technology Co ltd
Priority to CN202010623486.XA priority Critical patent/CN111815597B/en
Publication of CN111815597A publication Critical patent/CN111815597A/en
Application granted granted Critical
Publication of CN111815597B publication Critical patent/CN111815597B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0012—Biomedical image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30048—Heart; Cardiac

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Medical Informatics (AREA)
  • Geometry (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

The application relates to a left ventricle long and short axis tangent plane extraction method, a device, computer equipment and a storage medium based on CT images, wherein the method comprises the following steps: acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X; cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence; acquiring a left ventricle heart chamber maximum frame Y in the long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y; and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve. Compared with the prior art, the scheme can obtain other standard azimuth heart section images based on the cardiac CT image of a single layer, and can evaluate the function of the left ventricle more accurately and comprehensively.

Description

Left ventricle long and short axis tangent plane extraction method and device based on CT image, computer equipment and storage medium
Technical Field
The present application relates to the field of computer technologies, and in particular, to a method and an apparatus for extracting a long and short axis slice of a left ventricle based on a CT image, a computer device, and a storage medium.
Background
In clinical practice, examination of the heart mainly includes anatomy, wall thickness, perfusion, etc., but due to the different angles of deflection of the central axis of each person, the heart is often observed from different positions such as the short axis position (SA), the two-chamber heart (2CH), and the four-chamber heart (4CH) for better assessment of cardiac function. The functional status of the left ventricle is particularly important when assessing the function of the heart.
To evaluate the function of the left ventricle, it is generally necessary to display the whole segment of the left ventricle, wherein the most important long and short axes, two-chamber heart and four-chamber heart are the short axes, the short axis refers to the ventricular section perpendicular to the long axis of the left ventricle, and the information such as the heart chamber inner diameter and the ventricular wall thickness can be clearly observed on the short axis. The two-chamber heart of the left ventricle refers to a long-axis section which can simultaneously display the structures of the left ventricle and the left atrium, and can be used for observing information such as a mitral valve, the anterior wall and the posterior wall of the left ventricle and the like. The four-chamber heart can simultaneously display long-axis sections of left and right ventricles and left and right atria, and can be used for observing information such as a tricuspid valve, a septal wall of the left ventricle, a lateral wall and the like. Therefore, in order to fully evaluate the function of the left ventricle, the extraction of the long and short axis slices of the left ventricle is very important.
At present, long axis position and short axis position views of a left ventricle are obtained mainly by adjusting imaging parameters, the operation process is time-consuming, deviation is easily caused by interference of human subjective factors, and the requirements of modern clinical application cannot be met.
Disclosure of Invention
The application provides a method, a device, computer equipment and a storage medium for extracting a long-axis and short-axis tangent plane of a left ventricle based on a CT image, which are used for solving the technical problems that the long-axis tangent plane and the short-axis tangent plane of the left ventricle are determined in the prior art, the operation process is time-consuming, and the deviation is easily caused by the interference of human subjective factors.
A left ventricle long and short axis tangent plane extraction method based on CT images comprises the following steps:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in the long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
Optionally, the obtaining the apex point a 'and the mitral valve center point B' includes:
the line connecting the apex A 'and the center B' of the mitral valve is the long axis direction of the left ventricle.
Optionally, the obtaining the short-axis section sequence includes:
and acquiring a left ventricle heart chamber maximum frame Z in the short axis section sequence, marking a partition wall starting point and a partition wall ending point in the left ventricle heart chamber maximum frame Z, and calculating the vector average value from the starting point and the partition wall ending point to the center position of the left ventricle respectively to obtain the left ventricle short axis.
Optionally, the method for extracting the long and short axis slices of the left ventricle based on the CT image further includes:
and cutting the obtained short-axis section sequence to obtain a two-cavity core 2CH or a four-cavity core 4 CH.
Optionally, the method for extracting the long and short axis slices of the left ventricle based on the CT image further includes:
the short axis slice sequence comprises a plurality of left ventricle short axis images;
and inputting the left ventricle short axis image into a depth neural network model to obtain a boundary segmentation result of the left ventricle short axis image.
Optionally, the training path of the deep neural network model includes a contraction path and an expansion path:
the contraction path comprises 9 coding blocks, each coding block comprises two 3 x 3 convolution layers, a batch normalization layer and a ReLU activation layer are added behind each convolution layer, and downsampling is carried out on the first 8 coding blocks by using a 2 x 2 maximum pooling layer;
the expansion path comprises 8 decoding blocks, each decoding block comprises two 3 x 3 convolutional layers, a batch normalization layer and a ReLU activation layer are added behind each convolutional layer, and each decoding block performs upsampling by utilizing a 2 x 2 deconvolution layer;
and jump connection is carried out on the feature maps of the corresponding blocks in the contraction path and the expansion path according to the channel dimension.
Optionally, a residual structure is incorporated into each block of the deep neural network model; in the dilation path, the first 7 upsampling operations add an auxiliary path that serves as deep supervision and add an attention mechanism that gives different weights to different channels.
Optionally, the attention mechanism includes a channel domain attention mechanism and a spatial domain attention mechanism.
Optionally, the training image set of the neural network model is obtained in a manner that:
acquiring a plurality of left ventricle short axis images to form an initial image set;
and during each iteration in the training process, obtaining an enhanced image by randomly rotating, randomly zooming and randomly shearing each left ventricle short axis image in the initial image set.
Optionally, the random rotation angle range is-5 ° to 5 °, the random scaling range is 0.90 to 1.10, and the random shearing angle range is-5 ° to 5 °.
The application also provides the following technical scheme:
a left ventricle long and short axis tangent plane extraction device based on CT image includes:
the first module is used for acquiring a left ventricle heart chamber maximum frame X in the CT image and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
the second module is used for cutting the CT image in a manner of being parallel to a connecting line of the apex A and the center point B of the mitral valve and being vertical to the CT scanning plane to obtain a long-axis section sequence;
a third module, configured to obtain a left ventricular cardiac chamber maximum frame Y in the long-axis section sequence, and mark an apex point a 'and a mitral valve center point B' in the left ventricular cardiac chamber maximum frame Y;
and the fourth module is used for cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
The application also provides the following technical scheme:
a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in the long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
The application also provides the following technical scheme:
a computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in the long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
According to the method, the device, the computer equipment and the storage medium for extracting the long and short axis sections of the left ventricle based on the CT image, other standard position heart section images can be obtained based on the single-layer heart CT image, the function of the left ventricle can be evaluated more accurately and comprehensively, and the clinical requirements of a cardiologist for processing the heart CT image are met.
Drawings
FIG. 1 is a schematic flow chart illustrating a method for extracting long and short axis slices of the left ventricle based on CT images according to an embodiment;
FIG. 2 is a schematic diagram of a left ventricular chamber maximum frame X;
FIG. 3 is a schematic diagram of the left ventricular chamber maximum frame Y;
FIG. 4 is a schematic view of a short axis slice image;
FIG. 5 is a schematic structural diagram of a training path of a deep neural network model;
FIG. 6 is a schematic flow chart of a method of a deep neural network model;
FIG. 7 is a schematic diagram of a left ventricle segmented using the method provided by the present application and a prototype U-Net method; wherein:
(a-e) is a schematic diagram of the segmentation result of 5 graphs in the verification set;
FIG. 8 is a schematic three-dimensional model of a left ventricular wall in one embodiment;
FIG. 9 is a schematic representation of a three-dimensional model of a left ventricular chamber in one embodiment;
FIG. 10 is a schematic view of an arbitrary section of a three-dimensional model of the left ventricle;
FIG. 11 is a bullseye chart of wall thickening of the left ventricle;
FIG. 12 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The cardiac CT tomography is a single-slice scan, and can only obtain CT images in the same orientation of the heart, and the cardiac anatomy structure determines that the scanning direction often forms a certain included angle with the long axis of the left ventricle, so that the original CT image needs to be re-cut to obtain the standard orientation view of the left ventricle.
In order to extract the long and short axis slices of the left ventricle of the CT image, as shown in fig. 1, the method for extracting the long and short axis slices of the left ventricle based on the CT image includes the following steps:
step S100, acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
step S200, cutting the CT image in a manner of being parallel to a connecting line of the apex A and the center B of the mitral valve and being vertical to the CT scanning plane to obtain a long-axis section sequence;
step S300, acquiring a left ventricle heart chamber maximum frame Y in a long axis section sequence, and marking an apex A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and step S400, cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line between the apex A 'and the center point B' of the mitral valve.
The heart CT image processing method and the heart CT image processing device can obtain other standard position heart tangent plane images based on the heart CT image of a single layer, can evaluate the functions of the left ventricle more accurately and comprehensively, and meet the clinical requirements of cardiologists who process the heart CT image. Compared with CMR scanning, the scanning direction does not need to be adjusted in the scanning process, and the image cutting method is used for replacing multiple times of scanning, so that the difference between CT scanning and CMR scanning in the aspect of evaluating the cardiac function is greatly reduced.
If the extracted cutting long-axis section sequence and the extracted cutting short-axis section sequence are not satisfactory, the CT image can be cut again, and the cutting point and the cutting angle are corrected, so that the clinical requirements can be better met, and the use value of the cardiac CT image is improved.
The left ventricle heart chamber maximum frame X and the left ventricle heart chamber maximum frame Y can be selected manually, fig. 2 is a step of selecting the left ventricle heart chamber maximum frame X, and fig. 3 is a step of selecting the left ventricle heart chamber maximum frame Y;
in fig. 2 and 3, the apex point a, the mitral valve center point B, the apex point a ', and the mitral valve center point B' may be manually marked on the left ventricular chamber maximum frame X and the left ventricular chamber maximum frame Y, respectively.
The long-axis slice sequence includes a plurality of left ventricular long-axis images, and the short-axis slice sequence includes a plurality of left ventricular short-axis images.
It should be understood that, although the steps in the flowchart of fig. 1 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 1 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In another embodiment, as shown in fig. 3, obtaining the apex point a 'and the mitral valve center point B' comprises:
the line connecting the apex A 'and the center B' of the mitral valve is the long axis direction of the left ventricle.
In another embodiment, as shown in fig. 4, obtaining the short-axis slice sequence comprises:
obtaining a left ventricle heart chamber maximum frame Z in the short axis section sequence, marking a partition wall starting point and a partition wall ending point in the left ventricle heart chamber maximum frame Z, and calculating a radial average value from the starting point and the partition wall ending point to the heart chamber center of the heart chamber maximum frame Z to obtain a left ventricle short axis.
In fig. 4, the center position of the left ventricle is n, the start point is q, and the end point is p. The septal wall is located between the left ventricle and the right ventricle.
In another embodiment, the method for extracting long and short axis slices of the left ventricle based on the CT image further includes:
and cutting the obtained short-axis section sequence to obtain two-cavity core 2CH or four-cavity core 4 CH.
Cutting perpendicular to the short axis section sequence and parallel to the long axis to obtain two-chamber 2CH or four-chamber 4 CH.
In another embodiment, the method for extracting long and short axis slices of the left ventricle based on the CT image further includes:
and inputting the short-axis image of the left ventricle into the deep neural network model to obtain a boundary segmentation result of the short-axis image of the left ventricle.
The deep neural network model is obtained based on 8-layer coding-decoding structure training, can extract deeper abstract information, can realize rapid and automatic segmentation of left ventricle boundaries (inner and outer boundaries) in the short axis image, and can improve the accuracy of the left ventricle short axis image segmentation result.
In another embodiment, the training paths of the deep neural network model as shown in FIG. 5 include a systolic path and a dilated path:
the contraction path comprises 9 coding blocks, each coding block comprises two 3 x 3 convolution layers (step length is 1, zero padding number is 1), a batch normalization layer (namely a BN layer, network convergence acceleration) and a ReLU activation layer are added behind each convolution layer, a 2 x 2 maximum pooling layer (step length is 2, zero padding is not available) is used for downsampling behind the first 8 coding blocks, the size of a characteristic diagram is reduced by half, the number of characteristic channels is doubled to the first 3 x 3 convolution layer of the next coding block;
the expansion path comprises 8 decoding blocks, each decoding block comprises two 3 × 3 convolutional layers, a batch normalization layer and a ReLU activation layer are added behind each convolutional layer, a 2 × 2 deconvolution layer (step length is 2, zero filling does not exist) is used for up-sampling before each decoding block, the size of a feature diagram is doubled, and the number of feature channels is halved from the first 3 × 3 convolutional layer of the next decoding block;
and carrying out jump connection on the feature maps of the corresponding blocks in the contraction path and the expansion path according to the channel dimension.
In the last decoded block, the eigenvectors are mapped into one-dimensional space using a 1 × 1 convolutional layer (step size 1, no zero padding).
The contraction path is used for coding an input image and acquiring context information, and the expansion path is used for decoding and repairing the details and the space dimension of the image, generating pixel-level label output and realizing accurate positioning. In the positioning operation, a jump connection in the form of an overlapping operation is adopted, shallow features of each layer in the contraction path are connected with deep features of the corresponding layer in the expansion path, wherein the deep features are obtained through an upsampling operation, and then continuous convolution layers learn on the basis of the information, so that the problem of gradient dispersion caused by the increase of the number of network layers is relieved.
In another embodiment, in each block of the deep neural network model, a residual structure is incorporated; the input of the first 3 x 3 convolutional layer is added to the output of the second 3 x 3 convolutional layer to better optimize the deepened network.
The number of the characteristic channels of the 9 coding blocks is 8, 16, 32, 64, 128, 256, 512, 1024 and 2048 in sequence. The entire encoder section contains 27 convolutional layers in total.
In the dilation path, the previous 7 upsampling operations are added with auxiliary paths (7 auxiliary paths in fig. 5: aux 1-aux 7) used as deep supervision, and attention mechanisms for giving different weights to different channels are added.
And the deep characteristic diagram can be supervised by carrying out deep supervision through the auxiliary path so as to better restore the details and improve the segmentation precision. The current feature map is restored to the same size as the original image through a plurality of upsampling operations and 1 multiplied by 1 convolutional layer, the feature vectors are mapped to a one-dimensional space, and the loss values output by the 7 auxiliary paths and the loss value output by the last decoding block are weighted and summed from deep to shallow by 0.02, 0.04, 0.06, 0.08, 0.1, 0.2, 0.5 and 1 respectively.
In another embodiment, the attention mechanism includes a channel domain attention mechanism and a spatial domain attention mechanism.
Specifically, the basic information of the feature map is given as f, f is subjected to global average pooling, a 1 × 1 convolution layer and a Sigmoid activation layer (tool) are sequentially used to obtain the weight cw of a channel domain, and the output channel domain feature is fc ═ cw × f; and obtaining the weight sw of a spatial domain by sequentially using the 1 × 1 convolution layer and the Sigmoid active layer for fc, wherein the output spatial domain is characterized by fs being equal to sw × fc. Multiplying the features of different channels by different weights to enhance the attention of the key channel domain; features of different spaces are multiplied by different weights to enhance the attention of the key spatial domain. The channel domain attention mechanism directly and globally pools the features in the channel, spatial information in the channel is ignored, different weights can be given to the features in different regions in the channel by the spatial domain attention mechanism, the two weights are complementary, and the weights are automatically learned so as to better pay attention to the key features, and therefore the performance of the model is further improved.
The number of characteristic channels of the 8 decoding blocks is 1024, 512, 256, 128, 64, 32, 16 and 8 in sequence. The entire decoder section contains a total of 84 convolutional layers.
In another embodiment, the training image set of the neural network model is obtained by:
acquiring a plurality of left ventricle short axis images to form an initial image set;
during each iteration in the training process, each left ventricle short axis image in the initial image set is subjected to random rotation, random scaling and random shearing to obtain an enhanced image, so that the generalization capability and robustness of the neural network model are improved.
The initial image set and the augmented image together comprise a training image set.
In another embodiment, the angle of random rotation ranges from-5 to 5, the scale of random scaling ranges from 0.90 to 1.10 (less than 1 for magnification and more than 1 for reduction), and the angle of random shearing ranges from-5 to 5.
In another embodiment, as shown in fig. 6, the training process of the neural network model includes:
step 1, when deep neural network model training is carried out, weight initialization of the neural network is carried out by using He initialization, 800 training sets and 200 testing sets are read, and normalization operation of zero mean value and unit variance is carried out on the training sets and the testing sets. The learning rate is set to 2e-4, the block size is set to 8, and the iteration batch epochs is set to 200. Recording the current iteration batch epoch as 0, and entering the step 2;
step 2, performing data online enhancement on 800 training set images to improve the generalization and robustness of the network;
and 3, randomly dividing 800 training set images into 100 blocks (batch), wherein each batch comprises 8 images. Recording the current block batch as 0, and entering step 4;
step 4, inputting the current batch into the neural network model in the application;
and 5, estimating the inconsistency degree of the predicted value and the true value of the network model by adopting a Binary Cross Entropy (BCE) loss function, and updating the network parameters layer by using a Back Propagation (BP) algorithm through an Adam optimization function. Updating the batch to be batch +1, if the batch is less than 100, entering the step 4, otherwise, entering the step 6;
and 6, generating an alternative deep neural network model, inputting 200 verification sets for verification and evaluation, and if the evaluation result is better, storing the current alternative deep neural network model. Updating epoch to be epoch +1, if the epoch is less than 100, entering the step 2, otherwise, entering the step 7;
and 7, finishing the training, and saving the optimal alternative deep neural network model as a deep neural network model for segmenting the CT left ventricle short axis image.
DSC (formula (i)) measures the degree of spatial coincidence between the divided target region and the standard target region, Precision (Precision) (formula (ii)) measures the proportion of pixels that are actually positive in the divided target region, and Sensitivity (Sensitivity) (formula (iii)) measures the proportion of pixels that are correctly divided in the standard target region. The evaluation values of the three indexes are all between 0 and 1, and the larger the value is, the higher the consistency between the omega Seg and the omega GT is, the better the segmentation result is.
DSC＝2·(ΩSeg∩ΩGT)/(ΩSeg+ΩGT) Is 2. TP/(2. TP + FP + FN) formula (I)
Precision＝(ΩSeg∩ΩGT)/ΩSegTP/(TP + FP) formula (II)
Sensitivity＝(ΩSeg∩ΩGT)/ΩGTTP/(TP + FN) formula (III)
TABLE 1
Figure BDA0002563893470000101
Table 1 shows the evaluation results of the test set.
As can be seen from fig. 7 and table 1, for CT left ventricle short axis image segmentation, the method of the present application has higher segmentation accuracy and robustness compared to the prototype U-Net.
In another embodiment, as shown in fig. 8 and 9, the left ventricle three-dimensional reconstruction module superimposes the boundary segmentation results of the left ventricle short axis images according to three-dimensional coordinates to obtain a left ventricle three-dimensional model, and simultaneously, automatically calculates the left ventricle mass and volume, and then automatically calculates the left ventricle functional parameters such as cardiac output, cardiac index, ejection fraction, and the like according to the left ventricle volume at the end systole and end diastole.
Wherein, the cardiac output is the blood flow of the left ventricle per minute, the cardiac index is the cardiac output per unit body surface area, and the ejection fraction is the percentage of the stroke volume to the end-diastolic volume of the left ventricle.
The left ventricle three-dimensional model automatically calculates the volumes of the left ventricle myocardium and the heart cavity by using a Simpson method. In particular, it is assumed that the entire left ventricle consists of n short-axis slices P perpendicular to the long axis1，P2…PnThe composition is h in layer height, and is suitable for any short-axis section PiThe myocardial area S1 is automatically calculated by using a calculus methodiHeart chamber area S2i；
The calculation formula of the myocardial volume of the left ventricle is
Figure BDA0002563893470000111
The calculation formula of the heart cavity volume is as follows:
Figure BDA0002563893470000112
after the myocardial volume and the cardiac cavity volume of the left ventricle are calculated by the formula, the product of the myocardial volume and the myocardial density is the left ventricle mass. And then, by combining the definition of functional parameters of the left ventricle and utilizing the difference of the cardiac cavity volumes of the left ventricle at the end systole and the end diastole, the cardiac output, the cardiac index, the ejection fraction and the like can be automatically calculated.
In another embodiment, the bull's eye plot analysis module analyzes primarily left ventricular wall thickness and myocardial motion. The module has two functions, the first is arbitrary section view of the left ventricle three-dimensional model, as shown in fig. 10; the second function is bullseye map generation showing wall thickness information and intimal motion, as shown in fig. 11.
Based on the left ventricle three-dimensional model, the bull's eye graph analysis module provides a section browsing function, can manually adjust a cutting point and a cutting direction, automatically outputs a section view of the left ventricle on the cutting surface, and is beneficial to a clinician to selectively observe and detect a specific position of the left ventricle three-dimensional model.
According to the definition of a bull's eye diagram analysis module, the left ventricle three-dimensional model is divided into 4 parts of a base part, a middle part, an apex part and an apex, wherein the base part and the middle part are respectively divided into 6 blocks, the apex part is 4 blocks, and the apex is an independent block. After dividing the left ventricle into 17 corresponding blocks according to the definition of the bull's eye diagram, the wall thickness and the inner diameter value of each block except the apex of the heart are respectively calculated on the short-axis tangent plane. When the heart contracts, if the cardiac muscle of the left ventricle contracts normally, the wall thickness is increased, and the bull's eye diagram analysis module can display the information of the wall thickening of the left ventricle from the end diastole to the end systole. When the heart is in diastole, if the cardiac muscle of the left ventricle moves normally, the inner diameter is increased, and the bull's eye graph analysis module can also display the change information of the inner diameter of the left ventricle from the end systole to the end diastole.
In one embodiment, there is provided a left ventricle long and short axis slice extraction device based on CT image, including:
the first module is used for acquiring a left ventricle heart chamber maximum frame X in the CT image and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
the second module is used for cutting the CT image in a manner of being parallel to a connecting line of the apex A and the center point B of the mitral valve and being vertical to the CT scanning plane to obtain a long-axis section sequence;
a third module, configured to obtain a left ventricular cardiac chamber maximum frame Y in the long-axis section sequence, and mark an apex point a 'and a mitral valve center point B' in the left ventricular cardiac chamber maximum frame Y;
and the fourth module is used for cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the line is perpendicular to the line connecting the apex A 'and the center point B' of the mitral valve.
For specific limitations of the device for extracting the long and short axial slices of the left ventricle based on the CT image, reference may be made to the above limitations of the method for extracting the long and short axial slices of the left ventricle based on the CT image, and details are not described here. All or part of the modules in the device for extracting the long and short axial slices of the left ventricle based on the CT image can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 12. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to realize a method for extracting the long and short axis tangent plane of the left ventricle based on the CT image. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 12 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in a long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in a long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features. When technical features in different embodiments are represented in the same drawing, it can be seen that the drawing also discloses a combination of the embodiments concerned.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. The left ventricle long and short axis tangent plane extraction method based on the CT image is characterized by comprising the following steps:
acquiring a left ventricle heart chamber maximum frame X in a CT image, and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
cutting the CT image in a direction parallel to the connecting line of the apex A and the center B of the mitral valve and perpendicular to the CT scanning plane to obtain a long-axis section sequence;
acquiring a left ventricle heart chamber maximum frame Y in the long axis section sequence, and marking a cardiac apex point A 'and a mitral valve center point B' in the left ventricle heart chamber maximum frame Y;
and cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
2. The method of claim 1, wherein the obtaining the apex point a 'and the mitral valve center point B' comprises:
the line connecting the apex A 'and the center B' of the mitral valve is the long axis direction of the left ventricle.
3. The method of claim 1 or 2, wherein the obtaining the short-axis slice sequence comprises:
and acquiring a left ventricle heart chamber maximum frame Z in the short axis section sequence, marking a partition wall starting point and a partition wall ending point in the left ventricle heart chamber maximum frame Z, and calculating the vector average value from the starting point and the partition wall ending point to the center position of the left ventricle respectively to obtain the left ventricle short axis.
4. The method as claimed in claim 1, wherein the method further comprises:
and cutting the obtained short-axis section sequence to obtain a two-cavity core 2CH or a four-cavity core 4 CH.
5. The method as claimed in claim 1, wherein the method further comprises:
the short axis slice sequence comprises a plurality of left ventricle short axis images;
and inputting the left ventricle short axis image into a depth neural network model to obtain a boundary segmentation result of the left ventricle short axis image.
6. The method of claim 5, wherein the training paths of the deep neural network model include a contraction path and an expansion path:
the contraction path comprises 9 coding blocks, each coding block comprises two 3 x 3 convolution layers, a batch normalization layer and a ReLU activation layer are added behind each convolution layer, and downsampling is carried out on the first 8 coding blocks by using a 2 x 2 maximum pooling layer;
the expansion path comprises 8 decoding blocks, each decoding block comprises two 3 x 3 convolutional layers, a batch normalization layer and a ReLU activation layer are added behind each convolutional layer, and each decoding block performs upsampling by utilizing a 2 x 2 deconvolution layer;
and jump connection is carried out on the feature maps of the corresponding blocks in the contraction path and the expansion path according to the channel dimension.
7. The method of claim 6, wherein a residual structure is incorporated into each block of the deep neural network model; in the dilation path, the first 7 upsampling operations add an auxiliary path that serves as deep supervision and add an attention mechanism that gives different weights to different channels.
8. Left ventricle major and minor axis tangent plane extraction element based on CT image, its characterized in that includes:
the first module is used for acquiring a left ventricle heart chamber maximum frame X in the CT image and marking an apex point A and a mitral valve center point B in the left ventricle heart chamber maximum frame X;
the second module is used for cutting the CT image in a manner of being parallel to a connecting line of the apex A and the center point B of the mitral valve and being vertical to the CT scanning plane to obtain a long-axis section sequence;
a third module, configured to obtain a left ventricular cardiac chamber maximum frame Y in the long-axis section sequence, and mark an apex point a 'and a mitral valve center point B' in the left ventricular cardiac chamber maximum frame Y;
and the fourth module is used for cutting the long-axis section sequence to obtain a short-axis section sequence, wherein the long-axis section sequence is perpendicular to a connecting line of the apex A 'and the center point B' of the mitral valve.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
CN202010623486.XA 2020-07-01 2020-07-01 Left ventricle long-short axis section extraction method and device based on CT image, computer equipment and storage medium Active CN111815597B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010623486.XA CN111815597B (en) 2020-07-01 2020-07-01 Left ventricle long-short axis section extraction method and device based on CT image, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010623486.XA CN111815597B (en) 2020-07-01 2020-07-01 Left ventricle long-short axis section extraction method and device based on CT image, computer equipment and storage medium

Publications (2)

Publication Number Publication Date
CN111815597A true CN111815597A (en) 2020-10-23
CN111815597B CN111815597B (en) 2024-04-12

Family

ID=72856248

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010623486.XA Active CN111815597B (en) 2020-07-01 2020-07-01 Left ventricle long-short axis section extraction method and device based on CT image, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN111815597B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112308845A (en) * 2020-11-03 2021-02-02 赛诺威盛科技（北京）有限公司 Left ventricle segmentation method and device and electronic equipment
CN113469948A (en) * 2021-06-08 2021-10-01 北京安德医智科技有限公司 Left ventricle segment identification method and device, electronic equipment and storage medium
CN115272374A (en) * 2022-08-01 2022-11-01 北京安德医智科技有限公司 Ventricular segmentation method and device for cardiac magnetic resonance image
CN117934465A (en) * 2024-03-21 2024-04-26 赛诺威盛科技(北京)股份有限公司 TMVR parameter determination method, device and equipment
CN117958949A (en) * 2024-03-28 2024-05-03 天津市鹰泰利安康医疗科技有限责任公司 Atrial fibrillation radio frequency ablation method and system

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070253609A1 (en) * 2006-04-28 2007-11-01 Aben Jean-Paul M M Method, Apparatus and Computer Program Product for Automatic Segmenting of Cardiac Chambers
WO2008139344A1 (en) * 2007-05-10 2008-11-20 Koninklijke Philips Electronics N.V. Model-based spect heart orientation estimation
US20130259337A1 (en) * 2010-09-29 2013-10-03 Siemens Corporation Cardiac Chamber Volume Computation from Contours and Base Plane in Cardiac MR Cine Images
CN108564590A (en) * 2018-04-20 2018-09-21 上海理工大学 A kind of right ventricle multichannel chromatogram dividing method based on cardiac magnetic resonance film short axis images
CN108805913A (en) * 2018-05-14 2018-11-13 首都医科大学附属北京安贞医院 A kind of fusion method of coronary artery CT images and cardiac ultrasonic strain imaging
CN109377470A (en) * 2018-03-20 2019-02-22 任昊星 A kind of heart disease risk forecasting system
CN110335236A (en) * 2019-04-29 2019-10-15 上海依智医疗技术有限公司 Processing method, processing unit, processing system and the medium of cardiologic medical image
CN110335235A (en) * 2019-04-29 2019-10-15 上海依智医疗技术有限公司 Processing unit, processing system and the medium of cardiologic medical image
CN111161216A (en) * 2019-12-09 2020-05-15 杭州脉流科技有限公司 Intravascular ultrasound image processing method, device, equipment and storage medium based on deep learning

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070253609A1 (en) * 2006-04-28 2007-11-01 Aben Jean-Paul M M Method, Apparatus and Computer Program Product for Automatic Segmenting of Cardiac Chambers
WO2008139344A1 (en) * 2007-05-10 2008-11-20 Koninklijke Philips Electronics N.V. Model-based spect heart orientation estimation
US20130259337A1 (en) * 2010-09-29 2013-10-03 Siemens Corporation Cardiac Chamber Volume Computation from Contours and Base Plane in Cardiac MR Cine Images
CN109377470A (en) * 2018-03-20 2019-02-22 任昊星 A kind of heart disease risk forecasting system
CN108564590A (en) * 2018-04-20 2018-09-21 上海理工大学 A kind of right ventricle multichannel chromatogram dividing method based on cardiac magnetic resonance film short axis images
CN108805913A (en) * 2018-05-14 2018-11-13 首都医科大学附属北京安贞医院 A kind of fusion method of coronary artery CT images and cardiac ultrasonic strain imaging
CN110335236A (en) * 2019-04-29 2019-10-15 上海依智医疗技术有限公司 Processing method, processing unit, processing system and the medium of cardiologic medical image
CN110335235A (en) * 2019-04-29 2019-10-15 上海依智医疗技术有限公司 Processing unit, processing system and the medium of cardiologic medical image
CN111161216A (en) * 2019-12-09 2020-05-15 杭州脉流科技有限公司 Intravascular ultrasound image processing method, device, equipment and storage medium based on deep learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
段凤霞;何文;杜丽娟;隗冬梅;: "应用组织运动二尖瓣环位移自动追踪技术评价经皮冠状动脉介入治疗后左心室整体收缩功能改变", 中华医学超声杂志(电子版), no. 08 *
赵涛: "磁共振成像在肥厚性心肌病中的临床应用研究", 《中国优秀博士学位论文全文数据库信息辑》, no. 05 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112308845A (en) * 2020-11-03 2021-02-02 赛诺威盛科技（北京）有限公司 Left ventricle segmentation method and device and electronic equipment
CN113469948A (en) * 2021-06-08 2021-10-01 北京安德医智科技有限公司 Left ventricle segment identification method and device, electronic equipment and storage medium
CN115272374A (en) * 2022-08-01 2022-11-01 北京安德医智科技有限公司 Ventricular segmentation method and device for cardiac magnetic resonance image
CN117934465A (en) * 2024-03-21 2024-04-26 赛诺威盛科技(北京)股份有限公司 TMVR parameter determination method, device and equipment
CN117934465B (en) * 2024-03-21 2024-06-21 赛诺威盛科技(北京)股份有限公司 TMVR parameter determination method, device and equipment
CN117958949A (en) * 2024-03-28 2024-05-03 天津市鹰泰利安康医疗科技有限责任公司 Atrial fibrillation radio frequency ablation method and system
CN117958949B (en) * 2024-03-28 2024-05-28 天津市鹰泰利安康医疗科技有限责任公司 Atrial fibrillation radio frequency ablation simulation method and system

Also Published As

Publication number Publication date
CN111815597B (en) 2024-04-12

Similar Documents

Publication Publication Date Title
CN111815597B (en) Left ventricle long-short axis section extraction method and device based on CT image, computer equipment and storage medium
US10546378B2 (en) System and method of mitral valve quantification
CN111754534B (en) CT left ventricle short axis image segmentation method, device, computer equipment and storage medium based on deep neural network
US11551355B2 (en) Semi-automated heart valve morphometry and computational stress analysis from 3D images
Pouch et al. Semi-automated mitral valve morphometry and computational stress analysis using 3D ultrasound
US10417777B2 (en) Image processing apparatus, image processing method, and non-transitory computer-readable storage medium
CN110751187B (en) Training method of abnormal area image generation network and related product
CN113129418B (en) Target surface reconstruction method, device, equipment and medium based on three-dimensional image
CN117036530B (en) Cross-modal data-based coronary artery fractional flow reserve prediction method and device
WO2020007026A1 (en) Segmentation model training method and apparatus, and computer-readable storage medium
US11403761B2 (en) Probabilistic motion model for generating medical images or medical image sequences
US20140032180A1 (en) Method and apparatus for computing deformation of an object
CN111652954B (en) Left ventricle volume automatic calculation method, device, computer equipment and storage medium based on left ventricle segmentation picture
CN112365504A (en) CT left ventricle segmentation method, device, equipment and storage medium
CN110473208A (en) Structural form method of discrimination, computer equipment and storage medium
CN111816282A (en) Left ventricle functional analysis method and device based on CT image, computer equipment and storage medium
CN112767530B (en) Heart image three-dimensional reconstruction method, device, equipment and storage medium
US20230187052A1 (en) Automatic myocardial aneurysm assessment
CN111768391A (en) Full-automatic heart functional analysis method and device based on CT (computed tomography) image, computer equipment and storage medium
CN114847919A (en) Method for establishing patient-specific ventricular mathematical model for evaluating cardiovascular system based on nuclear magnetic resonance imaging
CN115578360A (en) Multi-target semantic segmentation method for ultrasonic cardiogram
CN115511908A (en) Medical image segmentation method, medical image segmentation device, computer equipment and storage medium
Garson et al. P2A-5 Guided Automatic Segmentation of the Murine Left Ventricle Using Conservation of Myocardial Volume
Qazi et al. Automated Heart Wall Motion Abnormality Detection using Sparse Linear Classifiers

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant